The authors consider a linear multiple objective decision making problem in a probabilistic framework. A novel scheme for solving such problems is proposed. This scheme consists of two phases: In phase I, the problem is optimized with the random parameters taking their mean values. With reference to the obtained Pareto solution, the decision makers choose critical target levels for the objects containing random parameters. In phase II, with those levels, an equivalent deterministic problem is formulated which takes into account the dispersion of the objectives and the eventual violation of the constraints in the face of different scenarios. This problem is solved by an interactive algorithm. A simplified regional development problem is solved by the developed scheme.< &gt;